What time is it? - a word about consistency and consensus in distributed systems

Thanks to the microservices concept applications can scale more than ever before. You can have multiple instances of the same service to distribute the data and decrease the load. What about data consistency? How to determine which service instance has the most actual version of the data?

In this talk, I would like to cover the data consistency in the microservices architecture problem. The talk is described in the below points:

1. Introduction -> what is the microservices concept? How does it help with heavy loaded systems?
2. Distributed systems mechanics ->how data is spread across different service instances? Explaining the concept of single-leader and multi-leader replication.
3. What can be the problem? -> How to deal with concurrent writes? What about “concurrent” reads (from different instances)?
4. What are the strategies to respond with consistent data? What are their pros and cons?

I am going to cover the following issues: - Linearizability - Ordering Guarantees - Distributed Transactions and Consensus